6.2 Reconstruction of the course of the voyage
6.2.2 Analysis of the electronically recorded track courses
6.2.2.4 Data analysis
6.2.2.4.1 Marginal conditions
A detailed analysis was carried out with the evaluation files produced (fused track data) using the SimDat software specially applied by Department of Maritime
25 The relevant information was kindly made available to the BSU by the Federal Maritime and Hydrographic Agency (BSH).
26 Cf. preceding footnote.
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
Navigation. All the results below are based on evaluations of these data. The following circumstances are to be taken into account as regards the precision and reliability of the data recorded:
• The position data recorded consist of GPS position data according to the characterisation in the data sets, which are less precise by comparison with DGPS position data.
• Differing degrees of precision of the dilution of precision delivered by GPS receivers can furthermore occur as a result of the configuration. For example, the dilution of precisions in the differing operating modes "Auto", "3D" or "2D dilution of precision" can differ. All the configuration settings carried out manually also influence the quality of the position, such as for example setting the S/N ratio higher or lower, admitting signals from satellites with very small angles of elevation or higher PDOP values27.
• The precision of the recorded positions can, moreover, deteriorate as a result of multiple reflections occurring on board, masking of the reception area or other ship-specific deflections.
• The position data of COSCO HAMBURG recorded are not direct GPS data. In the ECS system used, the so-called system position28 is recorded instead. Here the position data transmitted by the GPS receiver connected to the ECS are converted to the system position stipulated by the manufacturer. According to the information supplied by the manufacturer, the system position of COSCO HAMBURG regarding the midships axis is +/- 4.14 m (depending on the use of position sensor GPS1 or GPS2) and 210.9 m away from the stern.
Furthermore, the recorded position can be superimposed with an off-set that can be undertaken by the user as a manual adjustment.29 In the data analysis carried out in this respect, however, no data on the magnitude of any setting of such an adjustment carried out during the accident period could be found. That is why in this connection it is assumed that no manual adjustment is to be considered.
• The recorded position data of NEDLLOYD FINLAND are position measurements recorded directly by the receiver. Any off-set would be apparent in the recorded data in addition to this, as all ECS actions are recorded.30
• Any information from recorded data of the steering and propulsion facilities that might be relevant for the course of the accident were only available to a partial extent. While for COSCO HAMBURG the rudder angle, actual engine speed and furthermore actual rates of turn are stored, the ECS system used by NEDLLOYD
27 PDOP = Position dilution of precision.
28 The manufacturer has defined the system position as the position of the Doppler-Log oscillator in the vessel's bow.
29 Cf. Operating Instructions Chartpilot 9330; ECDIS, Conning Displays. SAM Electronics Hamburg, 10.07.2003.
30 Cf. Navi Sailor 3000 User Manual, TRANSAS Hamburg, May 2001.
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
FINLAND can only record engine speeds. In the case under consideration, however, the ECS system was not connected with the corresponding sensor.
• The synthetic radar data recorded by Vessel Traffic Services Brunsbüttel that were additionally made available to the Department of Maritime Navigation by the BSU could only be used, if at all, for a qualitative assessment in the meaning of a plausibility check of the position data recorded on board. On the one hand the synthetic data recorded are the position value extracted from the radar echo and calculated in accordance with a certain procedure (for example Leading Edge;
Center of Gravity or the like), whose relation to the actual ship's contour is not deterministic. On the other hand, the recording interval of 60 seconds is too large to support the reconstruction of the course of the accident.
6.2.2.4.2 Results of the raw data analysis
The data edited were analysed with the aid of the SimDat software. The results obtained from the analysis are set out below in diagrammatic form.
Figure 11: Example of evaluation of the situation analysis of the overtaking manoeuvre
Figure 11 shows the evaluation surface of the software used as an example. The diagrammatic representations in the right-hand part of the figure contain visualisations of the ECS data recorded.
In the upper diagram the tracks of the vessels involved (blue - COSCO HAMBURG;
red – NEDLLOYD FINLAND) and their true-to-scale vessel contours at the position at
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
the time 14:31:24 h together with selected chart objects (land contours (black), radar line (magenta) and navigation channel buoys) extracted from the ECS are shown in a cartesian X-Y-coordinate system (position data of the vessels transformed from Lat/Lon into Gauß/Krüger).
In the bottom diagram the speed tracks of the "speed over ground" values recorded for the two vessels are displayed over the entire analysis period observed. The green bars shown marks the time belonging to the situation set out above (here t= t0 = 0 seconds).
The following figures contain enlarged diagram representations.
Figure 12: Enlargement of excerpts from the visualised track courses at the start of the analysis period
Figure 13: Courses of speed over ground for the entire analysis period (blue – COSCO HAMBURG, red – NEDLLOYD FINLAND)
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
The speed tracks recorded show the approximation of the speeds to each other clearly. The speed of COSCO HAMBURG initially appears to be constant, while the speed of NEDLLOYD FINLAND initially drops slightly up to about t = 390 s (approx.
14.38 h) from some 13.5 kn to less than 11 kn, and then rises to almost 15 kn in approx. 50 s. After contact between the vessel hulls (probably time t = 490…505 s;
approx. 14.40 h), the speeds of the two vessels drop steeply.
In the following two figures the course tracks recorded through the water (Fig. 14 -heading) and over ground (Fig. 15) are shown. The heading of NEDLLOYD FINLAND is suddenly reduced shortly after t = 500 s from 130° to almost 90° and finally to 60°. At this time the vessel was lying almost crossways in the navigation channel.
Figure 14: Heading data
(green - NEDLLOYD FINLAND, red - COSCO HAMBURG)
Figure 15: Tracks of courses over ground (blue COSCO HAMBURG, red NEDLLOYD FINLAND)
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
The course over ground values recorded show that both vessels were largely proceeding on parallel courses. The course over ground of NEDLLOYD FINLAND only changes by almost 90° at the end of the period analysed. The additional off-set between the changes of the course values (over ground and heading) can be attributed to the fact that NEDLLOYD FINLAND, as evident on the photos of the accident, initially on still parallel courses approached COSCO HAMBURG and only turned crossways after touching the stern.
Figures 16 to 19 show further situation pictures of the overtaking manoeuvre on the basis of the data recorded.
Figure 16: Snapshot of the overtaking operation; closest approach of the sensor positions (Gauß-Krüger Coordinates)
Figure 17: Snapshot of the overtaking operation; closest approach of the sensor positions (geographical coordinates)
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
In figures 16 and 17 the location of the vessels in relation to each other roughly at the start of the overtaking is shown. At this time the sensor positions are roughly on the same height and the current distance determined between the vessels reaches its minimum value (see here also the diagram in Fig. 21). The distance between the vessel hulls reaches a first minimum at time t ≈ 90 s or approx. 14.33 h (see here also the diagram in Fig. 20). After this closest approach the sensor positions move ever further away from each other. At the probable time of collision (t ≈ 390...450 s;
14.40 h) the distance between the sensor positions is roughly constant and then increases continuously.
In order to calculate the relevant minimal distance from ship’s wall to ship’s wall, an additional function was implemented in the analysis software that determines the smallest distance by "spot" considerations from the side wall of one vessel to that of the other. On the one hand the corner points of the vessel contours themselves and on the other hand three further points along the side wall at 25%, 50% and 75% of the ship's length were used as "measuring points". All corner point combinations were calculated and evaluated to determine the minimal distance for all measurement times.
The diagrams in Fig. 18 and 19 show the tracks of the last evaluation segment from directly before to after the contact between the vessels. The positions marked by the ship's contours are faded in for the time of the second closest approximation.
Figure 18: Snapshot of the overtaking operation shortly before the collision (Gauß-Krüger Coordinates)
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
Figure 19: Snapshot of the overtaking operation shortly before the collision (geographical coordinates)
Figure 20: Closest distances between the vessel hulls of the vessels involved (minimal distance from ship’s wall to ship’s wall)
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
Figure 21: Closest distances between the sensor positions
(System position COSCO HAMBURG vs. GPS aerial NEDLLOYD FINLAND)
Figure 22: Absolute distances of the sensor position of COSCO HAMBURG to the radar line
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
Figure 23: Absolute distances of the sensor position of NEDLLOYD FINLAND to the radar line
It becomes evident from the data analysis conducted and the relevant figures that the contact between the two vessel hulls occurred close to the radar line. The recorded track of NEDLLOYD FINLAND shows a clear tendency towards the radar line and the track of COSCO HAMBURG. The situation of the collision can be reconstructed in good quality from the tracks and analysis of the location of the vessels (heading) in relation to each other. As the detailed analyses of the time course of the distance of the vessels to the radar line (Fig. 22 and 23) shows, COSCO HAMBURG especially had a clear lateral distance from the radar line in the last minutes prior to the collision and was running on a roughly parallel course, while the approach of NEDLLOYD FINLAND to the radar line took place almost continuously and without any course correction evident in the data sets.
However, the representations (cf. in particular Fig. 18 and 19) also make it clear that the actual progress of the situation cannot be reconstructed exactly with the aid of the data recorded without any further corrections. The contact of the vessel hulls clearly visible on the photos of the accident series could not be reproduced in the analysis of the recorded data.
On the basis of the photo series available, for example, it could be assumed that the relative approximation to each other is correct. In this case either the positions of COSCO HAMBURG on its own track must be deferred further backwards in time, or the positions of NEDLLOYD FINLAND must be pushed further forward in time.
Department of Maritime Navigation has carried out appropriate corrections by way of example. The following section contains explanations of these.
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
6.2.2.4.3 Subsequent time synchronising of the data sets
In order to be able to consider the probable course of the situation, Department of Maritime Navigation carried out synchronisation of the data sets on the basis of the photo series available. Figure 24 below shows a resulting snapshot shortly before the assumed time of the accident.
Figure 24: Presentation of the probable situation shortly before the occurrence of the accident (time-synchronised data sets)
Figure 24 shows the unchanged tracks of the two vessels and the hull contours for the time 14:39:24 in relation to the navigation channel limits and navigation marks taken from the BSH Database. While the position of NEDLLOYD FINLAND is still
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
taken from the original data sets, the position of COSCO HAMBURG was set back by 30 seconds. The basis for this additional synchronisation is the position of the vessels documented in Fig. 7 directly after the collision. According to this, the navigation mark "Pagensand-Nord Front-Lt" was almost dead ahead viewed from NEDLLOYD FINLAND shortly after the collision, while the stern of COSCO HAMBURG had already moved further away. As GPS data are recorded without further conversion on NEDLLOYD FINLAND, it was assumed that the position and time of this recording roughly correspond to the actual time of the accident.
Furthermore, it was assumed that the track of COSCO HAMBURG corresponds to the actual track, apart from the time allocation. It is further evident from the photos that the contact with the ship's bow of NEDLLOYD FINLAND occurred in the aft area roughly on a level with the last two 40-foot container bays. In order to reconstruct the collision, the data sets of COSCO HAMBURG were therefore set back in time in the available time steps recorded until the contact between the ships' hulls occurred. The further course of the situation reconstructed in this way is shown in the following figures.
Figure 25: Final phase of the contact between the vessels involved at approx. 14:39:34 h (time-sychronised data sets)
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
Figure 26: Situation shortly after the collision (14:39:44)
(time-synchronised data sets); NEDLLOYD FINLAND crossways in the navigation channel, navigation mark on the port side almost dead ahead)
The constellations shown in the last two figures are similar to the heading of the vessels to each other shown in the photo (Fig. 7). If this time sychronisation is taken as a basis for the further data evaluation, the distance developments set out below result.
By comparison with the distance developments determined from the data recorded (cf. Fig. 20; minimum value approx. 100 metres!), considering the distance from ship's wall to ship's wall after the time synchronisation results in a minimum value of approximately 0 m at the time of the assumed collision.
Figure 27: Closest distances from ship's wall to ship's wall (time-synchronised data sets)
Az.: 45 / 04
Bundesstelle für Seeunfalluntersuchung
Federal Bureau of Maritime Casualty Investigation
BSU
_____________________________________________________________________________________________________
Figure 28: Shortest distances between the sensor positions
(System position COSCO HAMBURG vs. GPS aerial NEDLLOYD FINLAND) time-synchronised data sets